This proposal describes four studies of physician behavior during their interactions with cancer patients. Our prior work in this area has disclosed that attitudinal and self-report scales do not correlate with actual physician behavior as measured on the Physician Behavior Check List (PBCL). These findings coupled with the fact that it is the behavior, per se, to which patients respond convinced us of the importance of studying actual behavior. The instrument we have developed to measure behavior (PBCL) is completed by a trained observer who accompanies physician groups on rounds; it allows both descriptive and quantitative analysis of physician behavior. Study 1 will examine the degree to which observer rated scores of the quality of the interaction and the actual behavior of the physicians correlate with actual patient satisfaction. Study 2 will measure the behavior of medical residents toward cancer patients and equally sick non-cancer patients to examine the widely held belief that cancer patients are treated differently. Study 3 will examine the PBCL and patient satisfaction instruments in an ambulatory office setting to allow for comparisons between inpatient and outpatient settings. Such comparisons will explore how the doctor-patient relationship changes as the severity of the illness increases. The ability to assess the effects of specific educational interventions on actual physician behavior would be a major advance in designing effective educational programs and in deciding how to allocate limited educational resources. Study 4 will directly test this question by using relatively simple instructional techniques (behavior modification) to attempt to change individual physician behavior associated with either poor observer ratings or low patient satisfaction. If we can improve physician behavior during the interaction with the cancer patient, we would expect improved doctor-patient relationships and an enhanced quality of life for cancer patients.